Are we building a bot here?

(Dibs on calling it the Seldon 9000!)

Resolves to "no" at the end of 2030 if no credible evidence of said training is available or earlier whenever a credible "yes" answer is verified. (We will have AGI hoovering up every extant bit of data available by that point anyway so the question is moot.)

Situations where Manifold user data was used without consent from the company (e.g. OpenAI using pirated ebooks to train ChatGPT) do not count.

The bot does not have to be released to the public/free to use/a standalone "Manifold" bot, there just has to be evidence that data produced by users here (predictions, comments, polls, account data like total profits, etc) were used in training some sort of AI that can do something similar to a user on Manifold can at the end of its training (e.g. it can make wildly inaccurate bets and lose all of its mana or post detailed questions that few or no people bet on). It doesn't have to be a highly effective predictor but ideally it should be able to at least simulate the behavior of the average Manifold user when reacting to new data (I.e. making reasonable predictions about future events based on data that did not exist when training).